In response to my post below, Michael Franz sends along the graph above and this note:
bq. I’ve been tinkering with a county-level model of presidential vote in 2008, looking for campaign effects of ads and visits. I include the number of solo visits by each of the candidates to the larger media market (Obama, McCain, Biden, and Palin). I count any joint appearances of Obama/Biden or McCain/Palin as “top of the ticket” visits. This is particularly key for the GOP, as McCain and Palin tended to campaign jointly for the first half of September. So the effect on Palin visits is for those where she campaigned alone. The graph shows the effect of visits holding constant all other factors, including ad buys and other visits by other candidates; the model was estimated with state-level fixed effects.
bq. The results are significant for Obama, McCain, and Palin, but not Biden. In fact, the slope of the Palin line is steeper than McCain’s, though Palin makes fewer visits. This ultimately only shows the effect on Election Day, so we miss any dynamics in September and October. But the results are pretty clear that Palin moved votes with her visits, and this is something that really hasn’t been demonstrated for more recent studies of visits by Lieberman and Cheney in 2000 and Edwards and Cheney in 2004.